242698 2022-07-13 10:18:26 -0700 REGRESSION: Colors are way off on http/tests/media/hls/hls-hdr-switch.html 2022-07-13 16:31:49 -0700 1 1 1 Unclassified WebKit Media WebKit Nightly Build Unspecified Unspecified NEW https://bugs.webkit.org/show_bug.cgi?id=242559 InRadar P2 Normal --- 1 ap webkit-unassigned eric.carlson heycam jer.noble webkit-bug-importer oldest_to_newest 1883143 0 ap 2022-07-13 10:18:26 -0700 http/tests/media/hls/hls-hdr-switch.html used to pass on macOS Big Sur as of 2021-04-15, before we broke it with PHP to Python conversion. Fixing that, I see that it fails in a new way. This test waits for a video drawn to canvas to become red, and then it waits for it to become green. The test allows for up to +-20 from (255, 0, 0, 255) and (0, 255, 0, 255) for those. But what I'm observing on my machine and on EWS too is (232,39,0,255) and (67,251,0,255). This is roughly the colors I see on the screen indeed, when running r-w-t- with --show-window. Getting these exact colors on Big Sur with both WK1 and WK2. As an aside, this test fails in a different way on macOS Ventura beta, I'm getting (0, 0, 0, 0). But this bug tracks the colors being off. 1883145 1 webkit-bug-importer 2022-07-13 10:18:37 -0700 <rdar://problem/96959572> 1883149 2 ap 2022-07-13 10:27:43 -0700 Making the tolerance higher in bug 242559. 1883445 3 ap 2022-07-13 16:31:49 -0700 I'm actually seeing exactly the same numbers on macOS Monterey (12.5). There was a regression on this major version introduced at a different time, but it must have been fixed.